Skip to NavigationSkip to content

The clinical data stream: locating the headwaters

Published on 02/06/09 at 08:14am

Way back in 1977, the US Food and Drug Administration (FDA) established a principle that would change clinical trials forever. Until then, a remarkable complacency had surrounded the subject for many years. Essentially, whenever a trial was published in a journal, it was assumed that the data underlying the results were valid. But the FDA had suspicions and carried out a major review, resulting in the publication of what we all now know as Good Clinical Practice (GCP). There were many components of this new standard, but the principle on which I will focus here is the data trail. The idea was - and still is - that whatever appears in the final report or publication should be traceable right back to original data. The definition of original data was the patient's medical records, and the process of source data verification (SDV) was thus firmly established.

The FDA's initiative was of course propagated around the world, and we now have an international standard, The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). Although this officially covers the major markets of Europe, North America and Japan, it is becoming a de facto standard worldwide. The data trail principle remains embedded within ICH GCP, despite major changes in technology and clinical practice over the last 30 years, and has to cope with huge geographical variations as well. This raises a wide range of practical issues.

As it happens, I have recently been (embarrassingly) racking up my air miles by scrutinising clinical research practices at centres in Europe, Asia and the Americas. This reminded me of all sorts of issues relating to SDV that I have encountered over the years, but one thing at least is clear: all the people involved know we must have SDV. What is intriguing is the range of methods that have evolved to enable it. Some of these make me wonder whether we are losing sight of why we need SDV.

To my simple mind, there is a temporal element to the linkage between medical records and clinical trial data. Surely the idea all along was that the investigator was to complete the medical records first, and then complete the case report form (CRF)? Well, we all live in the real world and we know that this doesn't always happen. For a start, GCP never expected that all the data in the CRF would be found in the medical records. The monitoring plan defines what the monitor should expect, and certain critical data points are always specified, such as visit dates, treatments, adverse events and main outcomes. Most of the remaining detail appears only in the CRF, so the CRF becomes the source. Results of investigations such as clinical pathology tests are printed automatically and again are the source (although I well remember the days when investigators had to transcribe them to the CRF).

But what if the monitor arrives at the site and finds key data points missing from the CRF? Nothing new in that of course, it happens all the time. If nothing is done there will be an audit finding, so the usual practice is to ask the investigator to make good the medical records. The problem with that is that it raises the question of what the source really is. The CRF was completed first, so it is in reality the source. Rivers rarely flow backwards. I have actually seen medical records completed after the CRF, but with earlier dates. Is this honest?

Such are the complexities of meeting GCP expectations in this regard, that many investigator sites have taken to designing their own data capture forms, which are placed in the patients' medical records as source data. Indeed, some contract research organisations supply these - they look almost identical to CRFs. The logic is clear, which is to ensure that the right data are in the medical records from the start, yet at the same time it can seem illogical. These sites are effectively writing down the same data twice. On reflection, it might be three times. Anything recorded for that patient before they went into the trial will already be in the records. It then goes on to the study-specific patient record, and then on to the CRF. Can anyone convince me that there will be no transcription errors?

Such study-specific records are also used to get around internal red tape surrounding access to centralised medical records. I know of a major site where it is so cumbersome to request a patient's notes that a member of the site staff goes to central medical records and copies data on to study-specific records. These are then maintained throughout the trial and consolidated with the central records at the end. A neat solution you might think, except what if another doctor sees the patient, say in the emergency department, and records other (non-trial) treatment? Indeed, the fragmentation of records can be a major challenge and is related to the movement of records between sites. For example, in India patients carry their own medical records from one hospital to the next, rather as health tourists, making centralisation and consolidation extremely difficult.

So let's try to bring this discussion back to the central issue, which is that of data integrity as defined by the regulators. We have seen that record keeping by doctors is often less than adequate for the needs of clinical research, triggering the appearance of study-specific documents. This is where the regulators can step in however. Anything created by the investigator for the study is regarded as distinct from the clinician's normal practice and is subject to the same quality assurance as every other study document. Sponsors have been criticised during regulatory inspections for not controlling such documents with the same rigour as they apply to everything else. Thus by seeking to 'ape' the CRF in patient notes, a rod is created for one's own back. Perhaps the answer lies in considering the questions that SDV was created to answer. These include:

* Does this patient exist? (Don't laugh, fraud does occur)

* Does the patient have the disease we are studying?

* Did the patient really attend for clinic visits?

* What treatment did the patient receive?

* How did the patient respond to treatment?

* Did anything go wrong?

If doctors are not recording such information, they are clearly being negligent and should not be doing clinical trials.

Les Rose is a freelance clinical science consultant and medical writer. Contact him on:

Mission Statement is a leading portal for the pharmaceutical industry, providing industry professionals with pharma news, pharma events, pharma service company listings and pharma jobs,
Site content is produced by our editorial team exclusively for and our industry newspaper Pharmafocus. Service company profiles and listings are taken from our pharmaceutical industry directory, Pharmafile, and presented in a unique Find and Compare format to ensure the most relevant matches